Transcript Service Multiplexing with Data-over

A Carrier’s Perspective on Implementing New
Data Protocols Over Legacy Networks
NFOEC 2003
September 10, 2003
Ian M. White1, Claudio Lima1, Greg Wolfe2, James Pan1
1Sprint Advanced Technology Laboratories
2Sprint Wireline Technology Development
[email protected]
Legacy Carriers and Next Generation
SONET/SDH
• Can legacy carriers benefit from the increasing demand for
data services, despite their architecture?
–The answer is YES.
– Next generation SONET/SDH will enable legacy carriers to costeffectively offer data services.
• Some technologies are already in place to use.
– Standards work is complete, products are on the horizon.
• Other technologies are immature or missing.
• We describe how legacy carriers can use Next Generation
SONET/SDH to deliver new data services.
– Benefits, cost savings, new services;
– Challenges, unmet requirements.
2
Legacy SONET/SDH Architecture
• SONET rings are the transport foundation of the metro core network.
• Some L2/L3 switching capability is available on metro switches.
• Many legacy IXCs reach clients through leased access.
– DS1/DS3 access is the most common (lack of deployed access fiber).
ILEC CO
ILEC
Metro/access
IP/MPLS/
ATM/FR
CPE
Co-located
Metro switch
CPE
IXC metro
core network
ILEC
Metro/access
IXC access methods:
leased DS-1/3,
OC-3/12, opt. Ethernet
OC48 rings.
SONET cross-connects with
L2/L3 switching capability.
OC192, ring or mesh.
IP/MPLS/ATM/FR core
3
Legacy Carriers Must Provide New Data
Services
• Must provide new data services using a legacy SONET/SDH network.
• This will require an overlay of new functions:
– Data over SONET/SDH, and data over PDH (DS1/3).
– Transport pipe size optimization.
– E-LAN support in the MAN and WAN.
– More flexible bandwidth control.
SAN
transport
IP/MPLS/
ATM/FR
E-LINE/ E-LAN
across the WAN
E-LINE/
E-LAN
4
Next Generation Data-Over-SONET/SDH
Protocols
• A new suite of protocols has been developed to enable legacy
carriers to provide data services with low investment.
– Generic Framing Procedure (GFP):
 Maps any data protocol into SONET/SDH.
 Transparent GFP (GFP-T) is optimized for Fiber Channel, FICON,
ESCON.
– Virtual Concatenation (VCat):
 Data can be mapped into any quantity of non-contiguous tributaries.
– Link Capacity Adjustment Scheme (LCAS):
 Standardized command set to enable hitless VCat capacity
modifications.
5
Advantages of GFP
• The literature has identified many advantages of GFP:
– Simpler, more robust packet delineation than HDLC.
– Handles header and payload bit errors.
– Maps any data protocol onto SONET/SDH.
• However, there will likely be only one immediately noticeable advantage to
carriers.
– Transparent GFP (GFP-T) transports Fiber Channel, FICON, ESCON over
SONET/SDH.
 This will enable legacy carriers to transport SAN data across any distance.
 This is likely to be a fruitful source of revenue because of the desire of the
financial industry for geographical separation between data centers and backup
sites.
• In the future, the ability to map any protocol may also be tremendously
beneficial to carriers.
– The Universal Line Card.
6
Advantages of VCat (1)
• VCat enables more efficient use of transport bandwidth.
– Can result in a significant savings on line card costs.
GigE
GigE
100MbE
100MbE
100MbE
IXC Metro
Ethernet Private Cross Connect
Line transport
in the metro
core network.
GigE
GigE
100MbE
100MbE
100MbE
OC-48
GigE
STS-48c
OC-48
GigE
STS-48c
OC-48
STS-3c
100MbE
Without VCat
OC-48
GigE
GigE
IXC Metro
Core Ring
IXC Metro
Cross Connect
STS-1-21v
STS-1-2v
100MbE
IXC Metro
Core Ring
With VCat
7
Advantages of VCat (2)
Ethernet Virtual Private Line
in the metro core network:
Transport Efficiency over SONET
Aggregated
Ethernet
STS-1-Nv
OC-48
Ethernet,
TDM
services
IXC Metro
Core Ring
With VCat
100
Efficiency (%)
L2
switch
120
80
60
Without VCat
40
20
0
0
500
1000
1500
2000
2500
Payload (Mb/s)
IXC Metro
Cross Connect
• Virtual concatenation can be used to efficiently size the transport pipe for
aggregated Ethernet traffic.
• Similarly, VCat can be used to efficiently size the transport pipe for RPR.
• With VCat, efficiency is better than 80%, and almost always better than 90%.
8
Low Order Virtual Concatenation
VT1.5-Nv
CPE
OC-3
10/100
Ethernet
To metro core
• LO VCat is NOT APPROPRIATE for the metro core and long-haul core network.
– To many circuits to manage (carrier operations teams and chip makers agree).
• However, LO VCat will be beneficial in the access area to transport Ethernet
from multi-tenant units (MTU).
9
VCat Differential Path Routing
• Can use differential path routing to create new level of service protection.
– Might be a long time before this fits into carriers’ operational models.
– Probably only useful in the long-haul core network.
• Could be problematic if deployed without LCAS (or similar function).
– Should quickly drop group members if their path is cut or severely errored.
• Differential delay compensation:
– ITU standard (G.707) suggests a maximum of 512 ms.
– In reality, anything over 100 ms is more than sufficient.
– In fact, 32 ms is sufficient for most continental networks.
metro
WAN
Diff. Delay
compensation
STS-1-2v
10
LCAS for Bandwidth Management and
Protection
• Standardized method of bandwidth adjustment is necessary.
– Bandwidth adjustment may occur across different metro areas with
different vendor equipment present.
• LCAS will be useful with VCat differential path routing if the data
service is not protected with SONET or RPR.
– If a path is lost or severely errored, the network management system
must quickly remove the affected VCat members from the group.
– Carriers may also use LCAS to add new members to restore the capacity.
Remove from group
after failure
metro
Diff. Delay
compensation
WAN
STS-1-2v
Re-add to group if
bandwidth is found
11
Service Multiplexing with Data-over-SONET/SDH
• New data services require service multiplexing:
– Aggregation of data being backhauled to the public Internet.
– Virtual Ethernet Private Line aggregation.
– E-LAN services.
• Only sites equipped with GFP/VCat can perform service multiplexing.
– GFP/VCat must be terminated before data can enter a switch.
• Service multiplexing for metro area services is straight forward.
– Metro switches are equipped with GFP/VCat and L2/L3 switching.
Service
multiplexing
E-LAN
E-LAN across
the WAN
IXC Metro
core network
IP/MPLS/
ATM/FR
IXC
Service
multiplexing WAN
Service
multiplexing
12
Service Multiplexing Across the WAN
• Service multiplexing across the WAN is less clear than the MAN.
– MPLS/ATM/IP will exist in the WAN, but GFP/VCat may not.
– To avoid GFP/VCat deployment in the core, carriers must avoid using
GFP/VCat with service multiplexing in the core.
 Terminate GFP/VCat at the POP at the edge of the core.
 Ethernet Private Line is still okay – no need to terminate GFP/VCat.
– This is probably an acceptable situation for carriers.
 Most carriers currently have excess bandwidth in their core.
 Thus, limiting GFP/VCat to the metro may be acceptable.
Terminate
GFP/VCat
Terminate
GFP/VCat
Terminate
GFP/VCat
metro
WAN
metro
IP/MPLS
IP/MPLS
IP/MPLS
13
Operations Management for Data Over SONET
• End-to-end operations management functionality for data over
SONET is immature.
• During GFP standard development, the expectation was to use the
GFP Client Management Frame (CMF) to carry Ops Mgmt data and
commands.
• This approach may have limitations.
– If a carriers wants to place centralized Ops Mgmt in the core, GFP/VCat
must be terminated (differential path routing constraint).
– Dial-up line to edge equipment from centralized management point may be
costly because many legacy carriers use leased access.
– Appears to be sufficient for Ops Mgmt of metro services.
14
Operations Management for EPL in the WAN
• Carrier may not want to use GFP CMF for Ops Mgmt of Ethernet
Private Line across the WAN.
• There is a proposal* in ITU to use SONET Path Overhead (POH).
– Tester is placed in the core of the network to monitor/intercept Ops Mgmt.
– Avoids the need to terminate GFP/VCat in the core.
– Avoids dial-up line access across leased access.
– Also works for architecture where GFP mapping is leased from wholesale
provider.
• Still in the early stages of development and definition.
*ITU-T Draft G.smc, version 0.0.1, January 2003
15
Data over PDH (DS1, DS3)
• Legacy IXCs spend a lot of money on local access.
– But costs must be minimized to cost-effectively support data services.
• Lowest cost approach: use DS1/DS3 to carry data into the metro.
– Use multiple DS1/DS3s bonded together to form the data pipe.
– Some initial offerings are available, but in general the technology is still
under development.
• Open question: Can VCat technology be leveraged for inverse
multiplexing data services across DS1/DS3?
• GFP over DS1/DS3 must be standardized.
– The standard is currently in progress.
16
Summary and Conclusions
• Legacy carriers will benefit from next generation dataover-SONET.
– VCat will save line card costs in the metro core.
 Potentially more than 50%, because of improvement in Gigabit
Ethernet transport.
– GFP will enable standardized transport of SAN data.
– GFP may eventually contribute to a universal line card.
– LCAS may offer interesting service protection options.
• More technology is needed.
– Definition of end-to-end operations management for data-overSONET.
– Cost-effective transport of Ethernet over DS-1/DS-3.
– Better understanding of how E-LAN functionality can be supported
across the WAN with/without GFP/VCat.
17